Colon cancer, and cancer in general, is a disease of the genome which invariably is associated with aberrant expression of genes. What is not clearly understood is how these genes are interconnected in the context of an operational network that underlies the cancer phenotype. An understanding of the network connections of cancer-related genes offers the potential to identify critical branch points amendable to therapeutic intervention. The goal of this application is to map cancer gene networks in a well-defined series of colon cancer cell lines from the NCI-60 panel. This will be accomplished by implementing an iterative process involving phenotyping, gene expression profiling, transcription factor binding analysis, and computational modeling of gene interactions. The resulting network model will be tested by perturbing cancer cell lines with molecular probes called short hairpin RNAs (siRNAs) in a process known as RNA interference (RNAi). Genes within the network whose knockdown by RNAi leads to a loss-of-cancer phenotype are hypothesized to regulate the expression of downstream effector genes. This perturbation in cancer cell lines will be examined by a subsequent round of gene expression profiling, transcription factor binding analysis and computational modeling, thereby allowing us to progressively improve our cancer network model. The validity of the revised model will be scrutinized with an ensuing round of RNAi perturbation and loss-of- cancer phenotype screening, and the iterative process is repeated. Lastly, the biological significance of genes in our networks will be cross-validated with expression profiling data of staged human colon cancers (normal colon, adenomas, Dukes'B, C and D stage samples, and liver metastases). Our research objectives are summarized as follows: (1) Construct a map of the invasion gene network, (2) Construct a map of the (anti-)adhesion gene network, and (3) Build probabilistic models of the invasion and (anti-) adhesion gene networks.